Dysregulation of intercellular signaling by MOF deletion leads to liver injury

Epigenetic mechanisms that alter heritable gene expression and chromatin structure play an essential role in many biological processes, including liver function. Human MOF (males absent on the first) is a histone acetyltransferase that is globally downregulated in human steatohepatitis. However, the function of MOF in the liver remains unclear. Here, we report that MOF plays an essential role in adult liver. Genetic deletion of Mof by Mx1-Cre in the liver leads to acute liver injury, with increase of lipid deposition and fibrosis akin to human steatohepatitis. Surprisingly, hepatocyte-specific Mof deletion had no overt liver abnormality. Using the in vitro coculturing experiment, we show that Mof deletion-induced liver injury requires coordinated changes and reciprocal signaling between hepatocytes and Kupffer cells, which enables feedforward regulation to augment inflammation and apoptotic responses. At the molecular level, Mof deletion induced characteristic changes in metabolic gene programs, which bore noticeable similarity to the molecular signature of human steatohepatitis. Simultaneous deletion of Mof in both hepatocytes and macrophages results in enhanced expression of inflammatory genes and NO signaling in vitro. These changes, in turn, lead to apoptosis of hepatocytes and lipotoxicity. Our work highlights the importance of histone acetyltransferase MOF in maintaining metabolic liver homeostasis and sheds light on the epigenetic dysregulation in liver pathogenesis.

Epigenetic mechanisms that alter heritable gene expression and chromatin structure play an essential role in many biological processes, including liver function. Human MOF (males absent on the first) is a histone acetyltransferase that is globally downregulated in human steatohepatitis. However, the function of MOF in the liver remains unclear. Here, we report that MOF plays an essential role in adult liver. Genetic deletion of Mof by Mx1-Cre in the liver leads to acute liver injury, with increase of lipid deposition and fibrosis akin to human steatohepatitis. Surprisingly, hepatocyte-specific Mof deletion had no overt liver abnormality. Using the in vitro coculturing experiment, we show that Mof deletion-induced liver injury requires coordinated changes and reciprocal signaling between hepatocytes and Kupffer cells, which enables feedforward regulation to augment inflammation and apoptotic responses. At the molecular level, Mof deletion induced characteristic changes in metabolic gene programs, which bore noticeable similarity to the molecular signature of human steatohepatitis. Simultaneous deletion of Mof in both hepatocytes and macrophages results in enhanced expression of inflammatory genes and NO signaling in vitro. These changes, in turn, lead to apoptosis of hepatocytes and lipotoxicity. Our work highlights the importance of histone acetyltransferase MOF in maintaining metabolic liver homeostasis and sheds light on the epigenetic dysregulation in liver pathogenesis.
Emerging evidence shows that epigenetic mechanism converts alterations in nutrient and metabolism into heritable patterns of gene expression and has profound implications in human physiology and diseases (1,2). Extensive interplays between epigenetic regulation and cell metabolism are reported to influence various cellular processes (3,4). For instance, the tricarboxylic acid (TCA) cycle generates by-products such as acetyl-CoA and S-adenosyl-methionine that are substrates of histone-modifying enzymes (3,5). Histone modifications, in turn, regulate expression of important metabolic genes that are critical for the catabolic and anabolic processes to support cell survival and growth. Histone modifications also directly modulate cell signaling to ensure the balance of nutrient availability and cellular capacity to use them effectively. Surprisingly, despite the prominent role of the liver in all metabolic processes in the body, there is a paucity of studies investigating deregulation of histone modifications and histone-modifying enzymes in the liver and their roles in common liver diseases. Among histone modifications, it is reported that global change of histone acetylation is associated with the progression of cirrhosis (6). Furthermore, altered expression or activity of the histone deacetylases (e.g., HDAC3) and sirtuins (e.g., SIRT1) is implicated in aberrant hepatic metabolism and progression of nonacoholic fatty liver disease (NAFLD) (7,8). These studies suggest that histone acetylation may play an important role in the liver. However, the physiological and pathological functions of histone acetyltransferases (HATs) have not been directly examined in the liver.
Among the histone acetyltransferases, males absent on the first (Mof, also called KAT8 or MYST1) is highly conserved and plays a nonredundant function in depositing lysine (K) 16 acetylation on histone H4 (H4K16ac) (9,10). H4K16ac is a prerequisite for additional H4 acetylation and higher-order chromatin structure and is associated with transcription activation (11,12). We and others have shown that Mof plays important function in embryonic stem cell (ESC) self-renewal (13,14), DNA damage repair (15,16), senescence (17), and autophagy (18). Mof also regulates fatty acid oxidation and mitochondria respiration. Mof deletion in ground-state ESCs leads to pluripotent quiescence by blocking fatty acid oxidation pathways (19). Mof depletion in cardiomyocytes increases reactive oxidative species (ROS) as a result of mitochondria dysregulation (20). In vivo studies show that deletion of Mof in the Mof f/f ; ER-Cre mouse model results in lethality in adult mice with postmortem liver abnormality (21). Significant reduction of MOF protein is found in a choline-and folatedeficient (CFD) mouse model of nonalcoholic steatohepatitis (NASH) (22). Lower level of H4K16ac is associated with poor overall disease-free survival of hepatocelluar carcinoma (HCC) patients (23,24). Despite these studies, the causal function of Mof in the liver has not been directly studied.
To examine the function of Mof in the liver, the main organ for metabolic processes in the body, we genetically deleted Mof in the liver. We find that simultaneous deletion of Mof, by Mx1-Cre, in multiple cell compartments in the liver leads to acute liver injury with increase of fat deposition, liver fibrosis, and cell death. Interestingly, the pleiotropic defects are not observed in mice with specific Mof deletion in hepatocytes. Mechanistic studies show that Mof deletion in both hepatocytes and Kupffer cells is necessary for liver pathogenesis and leads to feedback augmentation of inflammation and apoptosis signaling in the liver microenvironment. Consistent with importance of Mof in the liver, we find that MOF is frequently downregulated in human NASH patients and that Mofdependent gene program is often deregulated in this deadly liver disease (25,26). Taken together, our results show that deregulation of Mof is likely a novel contributor to metabolic liver diseases.

Establishment of the Mx1-Cre; Mof f/f mouse model
To study the function of MOF, we generated the Mx1-Cre; Mof f/f mouse model (Fig. S1A), which deletes Mof in multiple cellular compartments including hepatocytes and Kupffer cells (27,28). In this model, Mof was efficiently deleted at day 12 post poly-inosinic:poly-cytadilic acid (polyI:C) treatment ( Fig. 1A and Fig. S1A). Consistent with Mof deletion, both MOF protein and cognate histone H4 K16 acetylation (H4K16ac) were greatly reduced in the Mof −/− liver ( Fig. 1B and Fig. S1B). No Mof deletion was detected in the livers of the control polyI:C-treated Mof f/f or Mx1-Cre; Mof +/+ mice (Fig. S1A). To further confirm Mof deletion, we performed immunohistochemistry (IHC) for MOF in the livers of Mof f/f and Mx1-Cre; Mof f/f mice after polyI:C treatment. As shown in Figure 1C, MOF was largely absent in the nuclei of liver cells from the Mx1-Cre; Mof f/f mice. MOF was retained in a small subset of liver cells including endothelial cells and cholangiocytes surrounding veins and bile ducts as expected (Fig. 1C). Immunofluorescence for H4K16ac further confirmed loss of MOF in majority of liver cells (Fig. 1D). Since Mx1-Cre is also expressed in the hematopoietic system (27,29), we measured levels of lineage-committed hematopoietic cells in the peripheral blood and bone marrow upon Mof deletion. We did not observe significant changes in the levels of T cells (CD3 + ), B cells (B220 + ), or myeloid cells (Gr1 + , CD11b + ) in the peripheral blood (Fig. S1C). Similarly, no significant change in the hematopoietic progenitors was detected in the bone marrow for up to day 60 post polyI:C treatment ( Fig. S1C and data not shown).
Mof deletion leads to acute liver injury Upon polyI:C treatment, approximately 70% of Mx1-Cre; Mof f/f mice exhibited labored breathing and slowed gait in 3 weeks and had to be euthanized. The remaining Mx1-Cre; Mof f/f mice eventually succumbed within 60 days ( Fig. 2A). In contrast, the polyI:C-treated control Mx1-Cre; Mof f/+ , Mof +/+ and Mof f/f (no Cre) mice were normal at the study end point (80 days) ( Fig. 2A). Livers from moribund Mx1-Cre; Mof f/f mice were significantly enlarged as compared with those of the control mice (Fig. S2A). In some cases (2 out of 20), Mof −/− livers were visibly whitened due to significantly increased lipid deposition   Fig. S2B). Histology of the Mof −/− livers showed massive hepatic cell death radiating from the central vein (Fig. 2B), suggestive of terminal liver failure. Blood serum levels of liver enzymes, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), bilirubin (TBIL), and alkaline phosphatase (ALKP), were significantly elevated in Mof −/− mice as compared with the polyI:Ctreated control Mof f/f mice (Fig. 2, C-D). Since elevation of blood AST and ALT levels is often associated with hepatic steatosis and hepatitis (30), we examined the Mof −/− and Mof f/f livers for steatohepatitis-like features such as lipid deposition and fibrosis. Strikingly, majority of hepatocytes from Mof −/− mice showed enhanced accumulation of lipid microdroplets (Fig. 2E). Increase of Cyp2e1 expression, indicative of lipotoxicity, was also detected in Mof −/− livers (Fig. S2C). Furthermore, there was significant fibrosis in most Mof −/− livers isolated from moribund mice that survived past day 30 of polyI:C treatment (Fig. 2F, bottom). Taken together, Mof deletion by Mx1-Cre leads to severe liver injury, dysregulation of fatty acid metabolism, and increased liver fibrosis, which are characteristics of steatohepatitis (31).
MOF regulates gene pathways that are dysregulated in human liver disease.
To examine the function of Mof deletion at the molecular level, we performed RNA-seq analyses on primary liver tissues isolated from the Mof −/− and Mof f/f mice. There was a total of 1408 differentially regulated genes (fold change >2, p ≤ 0.05), of which 664 genes were upregulated and 744 genes were downregulated upon Mof deletion (Fig. 3A). Gene pathway analyses for the upregulated genes showed that they were enriched for pathways such as negative regulation of cell proliferation, apoptotic process, response to lipopolysaccharide, cytokine response, and fibroblast proliferation (Fig. 3B), consistent with steatohepatitis-like phenotypes in vivo (Fig. 2). Interestingly, the enriched gene pathways for the downregulated genes almost exclusively involve metabolic processes, such as oxidation reduction and the sterol and cholesterol biosynthetic pathways ( Fig. 3C) (Table S1). GSEA analysis further confirmed enrichment of apoptosis and inflammatory response pathways in the upregulated genes ( Fig. S3A), as well as enrichment of fatty acid metabolism and other mitochondria processes in the downregulated genes ( Fig. S3B). Heatmap of representative Mof targets in these pathways was shown in Figure 3D and expression of selected genes was confirmed by real-time PCR (Fig. S3C).
Given the steatohepatitis-like features in Mof null mice (e.g., increase of fat deposition, fibrosis, elevated serum AST and ALT), we compared gene pathways dysregulated by Mof deletion with those in human NASH patients. Interestingly, RNA-seq analyses of primary liver samples from human NASH patients and healthy controls (GSE134422) showed that MOF was significantly downregulated in human NASH patients (Fig. 4A). Furthermore, there were over 2283 genes with altered expression in human NASH patients. Among them, a significant subset of genes (12%) were dysregulated in both human NASH and the Mof deletion mouse model (Fig. 4, B-C). Majority (62.50%) of the commonly deregulated genes were involved in inflammatory response and apoptotic pathways in KEGG pathway analysis (Fig. 4D). Pathways such as oxidationreduction and lipoprotein metabolic pathways were among the commonly downregulated pathways in both human NASH patients and Mof null mice (Fig. 4E, Table S2). Expression of representative inflammatory signaling factors such as NOS2, CXCL5, and CCL2 and metabolic genes such as ACSS2, MMAB, and PIK3C2G in Mof null liver and human NASH patients were shown in Figure 4F. Similar conclusions could also be drawn from analyzing an independent RNA-seq data set from 16 human NASH patients (Fig. S4, B-D and Table S3) (31). These results suggest that Mof deletion-mediated transcriptome changes carry some molecular characteristics of human NASH.

Hepatocyte-specific Mof deletion had no apparent liver defects
Since Mx1-Cre is expressed in both hepatocytes and Kupffer cells in the liver (28), we decided to examine whether liver failure upon Mof deletion is intrinsic to hepatocytes. To this end, we specifically deleted Mof in hepatocytes of the Mof f/f mice by tail vein injection of adeno-associated virus (AAV) expressing Cre recombinase under the control of the promoter of hepatocyte-specific thyroxine-binding globulin gene (TBG) (AAV-TBG-Cre) (Fig. 5A). In parallel, we injected AAV expressing TBG-driven green fluorescent protein (AAV-TBG-GFP) as the negative control (Fig. 5A). Injection of AAV-TBG-Cre, but not AAV-TBG-GFP, specifically deleted Mof in the liver, but not in hematopoietic organs such as the spleen and peripheral blood, as indicated by the genotyping result for Mof excision (Fig. 5B) (32,33). Western blot analysis further confirmed significant reduction of both MOF protein and H4K16ac in the livers after AAV-TBG-Cre injection (Fig. 5C). The remanent signals for MOF and H4K16ac were likely from nonhepatocytes (e.g., Kupffer cells) in the liver, which maintained normal Mof expression. Surprisingly, unlike the Mx1-Cre mouse model, all mice were viable at least 3 months after AAV-TBG-Cre injection. Livers from mice injected with AAV-TBG-CRE or AAV-TBG-GFP showed no obvious difference. H&E staining showed normal liver architecture in Mof −/− mice (Fig. 5E). Consistently, serum level of ALT was normal at day 30 post deletion ( Fig. 5F). No fibrosis was detected in the liver of Mof −/− mice (Fig. 5G). These results suggest that hepatocyte-specific Mof deletion does not have detrimental effects in the liver.

Simultaneous deletion of Mof in hepatocytes and bone marrow-derived macrophages (BMDMs) increases apoptosis of hepatocytes in vitro
The discrepancy of Mof function in the Mx1-Cre and AAV-TBG-Cre mouse models suggests that Mof deletion-induced liver injury and steatohepatitis-like features probably require coordinated changes in both hepatocytes and Kupffer cells. To test this, we performed the in vitro coculture experiment using primary Mof −/− and Mof f/f hepatocytes with bone marrowderived macrophages (BMDMs) isolated from ER-Cre; Mof f/f mice (Fig. 6A). The BMDM are commonly used to study the function of Kupffer cells in vitro (34). Mof deletion in the BMDM was induced by adding 100 nM tamoxifen (4-OHT) to the cell culture 72 h prior to the experiment ( Fig. 6A and Fig. S5). Ethanol was used as the control for mock treatment. Simultaneous deletion of Mof in both hepatocytes and BMDM led to significant increase of apoptosis in hepatocytes (Fig. 6B, right), which is in contrast to that of Mof f/f hepatocytes cocultured with Mof −/− BMDM (Fig. 6B, left) or that of Mof −/− hepatocytes cocultured with mock-treated Mof f/f BMDM (Fig. 6B). Quantifications of the apoptosis analyses were shown in Figure 6C. Consistent with Annexin V/PI staining, the levels of cleaved PARP as well as active Caspase 3 proteins were significantly higher in Mof −/− hepatocytes after coculturing with Mof −/− BMDM than those after coculturing with mock-treated Mof f/f BMDM (Fig. 6D). Furthermore, cytosolic cytochrome c, a downstream effector of mitochondria apoptosis, was also significantly increased in Mof −/− hepatocytes after coculturing with Mof −/− BMDM, but not with mock-treated Mof f/f BMDM or no BMDM (Fig. 6D). Reciprocal changes of mitochondrial cytochrome c provided further confirmation (Fig. 6E). These results argue that liver injury observed in the Mx1-Cre; Mof f/f mice likely requires simultaneous loss of Mof in both cellular compartments in vivo.

Reciprocal signaling between hepatocytes and BMDM is required for inflammation response
To examine Mof-dependent signaling in the liver microenvironment that may contribute to liver injury, we first examined whether Mof −/− hepatocytes promote inflammation signaling to BMDM. To this end, we cultured the Mof −/− and Mof f/f BMDMs in the Mof −/− hepatocyte-conditioned medium. The Mof −/− BMDMs were activated by the Mof −/− hepatocyte-conditioned medium, expressing much higher level of chemokines CCL2, proinflammatory cytokine IL6, tumor necrosis factor (TNF)-α, profibrotic gene TIMP, and inducible nitric oxide synthase (iNOS) (Fig. 7A), consistent with in vivo RNA-seq analysis (Fig. 3D). Western blot analyses further confirmed changes of these genes at the protein levels  (45). B, Venn diagram of differentially expressed genes (fold change > 2) in Mof null mouse livers and human NASH samples. C, pie chart for genes dysregulated in both mouse and human datasets. D, upregulated biological processes in the human NASH dataset. X-axis shows negative log 10 p value. E, downregulated biological processes in human NASH dataset. X-axis shows negative log 10 p value. F, heatmap of expression of representative Mof targets in human NASH and healthy control samples as indicated on top. Heatmap key on bottom indicates normalized log 10 fold change. (Fig. 7B). In contrast, Mof −/− BMDM or Mof f/f BMDMs cultured in regular medium had no or only modest change in CCL2, IL6, or TNFα at gene expression (Fig. 7C) and protein levels (Fig. 7D). Importantly, heat treatment of the Mof −/− hepatocyte-conditioned medium abolished its ability to activate Mof −/− BMDM (Fig. 7E). These results suggest that Mof null hepatocyte may provide cytokine signaling for macrophage activation.
We next examined the reciprocal signaling from macrophages to hepatocytes. Mof −/− BMDM expressed higher level of iNOS (Fig. 7, C-D), which increases release of nitric oxide (NO) to trigger apoptotic response (35, 36). We next asked whether Mof −/− hepatocytes are more sensitive to NO signaling and contribute to a feedback amplification of cell death signaling. Since primary BMDMs are difficult to transfect and have short viability in vitro for genetic studies, we directly treated Mof f/f and Mof −/− hepatocytes with sodium nitroprusside (SNP) as the NO donor. As shown in Figure 7F, although 1 mM SNP treatment decreased ATP production in both Mof f/f and Mof −/− hepatocytes, Mof −/− hepatocytes were much more sensitive to NO signaling than Mof f/f hepatocytes (Fig. 7F). Mitochondria in 1 mM SNP-treated Mof −/− hepatocytes were significantly smaller and more punctuated as compared with the mock-treated cells (Fig. 7G), indicative of onset of apoptosis (37,38). Consistently, 1 mM SNP-treated Mof −/− cells released more cytochrome c from mitochondria into cytosol than that of the Mof f/f hepatocytes (Fig. 7H). These results suggest that Mof deletion probably altered the reciprocal signaling between hepatocytes and macrophages in the liver microenvironment, which leads to a feedforward amplification of the inflammation (BMDM) and cell death (hepatocyte) responses leading to liver injury and steatohepatitis-like features (see Discussion).

Discussion
Here we find that Mof deletion by Mx1-Cre leads to severe liver injury with increasing lipid deposition and fibrosis. Notably, the liver injury and inflammation upon Mof deletion bear some similarity with human steatohepatitis at both phenotypical and molecular levels. Manifestation of liver injury in Mof null mice is heterogeneous. Majority of mice have accumulation of small lipid droplets in the liver, and a small subset of Mof null mice develop apparent fatty liver diseases. Significant heterogeneity of the nonacoholic fatty liver spectrum has also been widely reported in human patients (39,40). About 20% patients have nonacoholic fatty liver or steatohepatitis without obesity or high-fat diet (39). It suggests that a multitude of factors (e.g., diet, genetic and epigenetic factors) may contribute to the disease progression. Comparing with the natural progression of NAFLD/NASH in patients and the widely used high-fat diet mouse model, Mof null mice acutely develop steatohepatitis-like liver injuries. It is likely that downregulation of MOF is one of the key downstream events in progression of these deadly liver diseases. In this scenario, MOF downregulation may modulate the cellular epigenetic landscape to activate a feedback loop that leads to sustained inflammation and eventual liver injury. Indeed, we find significant downregulation of MOF in human terminal NASH patients as well as aberrant expression of a common subset of genes in both human NASH and the Mof null mouse models. To our knowledge, our study is the first one to demonstrate a causal role of a histone acetyltransferase in liver abnormalities.
A previous study reported that hematopoietic stem cells isolated from Mx1-Cre; Mof f/f mice were unable to sustain long-term hematopoiesis after transplantation into the recipient mice (41). In our study, we did not observe any hematopoietic defects in the primary mice for the duration of our study. In contrast to liver injury observed in the Mx1-Cre; Mof f/f mouse model, hepatocyte-specific deletion by AAV-TBG-Cre has no overt defects in adult livers. Similarly, myeloid-specific Mof deletion by Lyz2-Cre does not affect macrophage or hepatic functions in mice (42). Thus, severe liver injury in the Mx1-Cre is likely due to simultaneous deletion of Mof in both hepatocytes and Kupffer cells in liver. This is supported by the in vitro coculture experiments using hepatocytes and BMDMs. We have revealed a causal role of Mof in regulating the reciprocal signaling between BMDM and hepatocytes. Mof deletion in both BMDM and hepatocytes triggers an aberrant proinflammatory cascade that is not observed in its deletion in either cell compartment alone. Interestingly, Mof deletion in hepatocytes increases cytokine signaling for BMDM activation, as exemplified by increase of TNF-α and toll-like receptor (TLR) signaling (e.g., TNFα, IL-6) (Fig. 7A). These signaling pathways have been reported as the major contributors to NASH progression in patients (43). The activated macrophages, in turn, release NO and other cytokines to induce apoptosis as well as proinflammatory responses in the liver as previously reported (44,45). In the feedback loop, Mof deletion in hepatocytes also enhances sensitivity to NO-mediated death signaling (Fig. 7). Thus, our study highlights the necessity of coordinated epigenetic dysregulation in both hepatocytes and Kupffer cells during development of the steatohepatitis-like liver injury.
By examining genes that are dysregulated in both Mof null mice, we reveal that MOF is important for regulating multiple metabolic pathways, including lipid metabolism and oxidation-reduction process. The metabolic aberration likely leads to oxidative stress in the liver microenvironment, which further disrupts hepatic lipid and cholesterol synthesis, perpetuating a feedback loop that aggravates liver dysfunction (46). The lipotoxic and oxidative stress are able to trigger a cascade of proinflammatory events. Both metabolic dysregulation and inflammation are major contributors to NASH progression in patients (43). Notably, previous studies have shown that hepatocyte-specific deletion of histone deacetylase Immunoblot for β-actin was used as the loading control. C, relative gene expression in Mof f/f BMDM after 100 mM 4OHT or EtOH treatment. Y-axis is fold change after normalization against Gapdh level, which was arbitrarily set as 1. Mean and standard deviation (error bar) from three independent experiments were shown. (***p < 0.001, two-way ANOVA test). D, immunoblot for CCL2, IL6, TNFα, INOS, and TIMP in Mof f/f BMDM after 100 mM 4OHT or EtOH treatment. Immunoblot for β-actin was included as the loading control. E, relative gene expression in Mof −/− BMDM cultured in L15 media, the conditioned medium from Mof −/− hepatocytes, or the conditioned medium after heating at 100 C for 5 min. Y-axis is fold change after normalization against expression of Gapdh, which is arbitrarily set as 1. Mean and standard deviation (error bar) from three independent experiments were shown. (*p < 0.05, ***p < 0.001, two-way ANOVA test). F, cell-titer Glo assay for ATP production at 0 or 20 h after SNP treatment in primary Mof −/− and Mof f/f hepatocytes. Mean and standard deviation (error bar) from three independent experiments were presented. (***p < 0.001, two-way ANOVA test). G, representative MitoTracker live staining of Mof −/− hepatocytes with or without 1 mM SNP treatment, scale bar 0.5 μm. Right panels were enlargement of the square area from the left images (scale bar 0.15 μm). H, western blots for cytosolic or mitochondria cytochrome c in Mof −/− hepatocytes with or without SNP treatment. Antibodies were indicated on the right.
Hdac3 and Sirt1 leads to steatosis (47,48). These histone deacetylases disrupt lipid and glucose homeostasis and reroute metabolic precursors toward lipid synthesis and lipid sequestration in vivo (47,48). Our finding that Mof is also important to regulate lipid homeostasis, through modulating fatty acid oxidation (e.g., Acss2, Acss3) and the cellular redox state, shows that the balance of histone acetylome is probably necessary to maintain metabolic homeostasis in the liver. Breaking the balance by depleting either histone acetyltransferase Mof or histone deacetylases will result in disruption of normal liver functions. Finally, it is worth noting that in addition to aggressive liver injury in the Mx1-Cre; Mof f/f mice, the gene pathway associated with cancer was also modestly enriched in Mof null liver and NASH patients (Fig. S4A). Given global downregulation of MOF and H4K16ac in HCC (23,24), it would be of interest to examine whether MOF plays a role in the progression of NASH to HCC in future.

Mouse strains and genotyping strategies
Generation of Mof f/f and Mof f/f ; ER-Cre alleles were previously described (13). The Mx1-Cre; Mof f/f mice were generated by breeding the Mof f/f mice to B6.Cg-Tg(Mx1-cre) 1Cgn/J mice (Jackson Laboratories, 003556). For Mof deletion in the Mx1-Cre model, polyI:C (Amersham) was intraperitoneally injected into mice at 2.5 μg/g concentration every other day for six consecutive doses. To generate hepatocyte-specific Mof deletion, 10 9 pfu of either AAV-TBG-GFP (control) or AAV-TBG-Cre viruses (to ablate hepatic floxed genes) in 100 μl of sterile PBS were injected into tail vein of the 8-weekold Mof f/f mice. Gene deletion in the liver of the AAV-TBG-Cre-injected mice was achieved after 2 weeks. Genotyping strategies were previously described (15) and illustrated in Fig. S1A and Figure 5B. Briefly, mouse tails (5-10 mg) were boiled in 150 μl NaOH (50 mM) for 25 min, followed by addition of 15 μl Tris-HCl (1 M, pH 7.5). Genomic DNA was used as the template for PCR reaction using primers (F: TGCTCGTGGTAGTTGACAGC, R: TGGGCTCCAGGA TAAACTTG). The reaction was carried out using cycling parameters: 94 C for 2 min; 35 cycles of 94 C for 30 s, 59 C for 30 s and 72 C for 30 s; followed by 72 C for 2 min. Using this method, successful Mof deletion can be detected as an 850 kb band on the agarose gel (15). All animal experiments were performed in accordance with guidelines set by the Institutional Animal Care and Use Committee (IACUC) at the University of Michigan and University of Southern California. Animal experimentations were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. In all experiments, equal ratio of male and female mice was used, and Mof deletion does not confer sex bias in our study.

Western blot analysis
Frozen liver tissues were ground in a mortar and pestle and lysed in the RIPA buffer (150 mM NaCl, 50 mmol/l Tris-HCl, pH 7.4, 1 mM EDTA, 1% Triton X-100 (Sigma-Aldrich), 1% sodium deoxycholic acid, 0.1% sodium dodecyl sulfate, 1 mM phenylmethylsulfonyl fluoride, and 1X cOmplete protease inhibitor cocktail (Thermo Fisher)). Primary hepatocytes were lysed in the RIPA buffer directly in the culture dish. Mitochondria were isolated from cells using the Mitochondria Isolation Kit (Ther-moFisher, 89874). Protein concentrations were determined by the Bradford assay (BioRad) and analyzed on an Ultraspec 2100 Pro spectrophotometer (Amersham Biosciences) at 595 nm. Five micrograms of total protein was loaded on 4% to 20% Mini-Protean TGX Precast gels (BioRad) and transferred to membranes using the Trans-Blot Cell system (BioRad). The blots were probed with following primary antibodies: anti-α-Tubulin

Hematoxylin and eosin, Masson's trichrome, and Sirius Red staining
For hematoxylin and eosin (H&E) or Masson's trichrome staining, livers were immersion-fixed with 10% buffered formalin and embedded in paraffin for sectioning. For the H&E staining, the deparaffinized slides were incubated by hematoxylin for 1 min followed by washing in flowing tap water for 5 min. The slides were then stained with eosin (Sigma HT110232) for 1 min, followed by wash with tap water and differentiation procedure. For the Masson's trichrome staining, liver paraffin slides were subjected to Masson's trichrome stain according to the manufacturer's instructions (Trichrome Stain Masson Kit; Sigma-Aldrich; HT15). For the Sirius Red staining, the mouse liver paraffin slides were baked at 60 C for 1 h and soaked in xylene and graded ethanol solutions (100%, 95%, 85%, 75%, 60%, 50% till 0%). Slides were then stained with 0.1% Sirius Red (Sigma, 365548) and 0.1% Fast Green (Sigma, F7252) (dissolved in saturated picric acid) overnight. The slides were washed with 10 mM hydrochloric acid for 2 min, rapidly dehydrated through graded alcohols starting at 70% and sealed with cover slips by Permount mounting medium.

Immunofluorescence and immunohistochemistry
Livers were embedded with Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA) and snap-freezed in liquid nitrogen. Cryostat sections were mounted on salinized slides and fixed with ice-cold acetone. Immunohistochemistry (IHC) for MOF was performed on cryostat sections using anti-MOF antibody from Abcam (ab2000660, 1:100 dilution). For immunofluorescence and Oil Red O staining, cryostat sections were mounted on salinized slides and fixed with ice-cold acetone. Fc receptors were blocked with 1% anti-mouse CD16/ 32 antibody (Biolegend, 10135) in 2% normal goat serum (NGS) for 30 min at room temperature. Slides were incubated with anti-Keratin 8 antibody (Lifespan Biosciences, LS-B7928), antiacetylated histone H4K16 (Abcam, ab109463), followed by Alexa-labeled secondary antibody (Invitrogen) at 1:200 dilution. Images were obtained using Olympus BX43 microscope and Cell Sens Software.

Blood chemistry
For the Mx1-Cre mouse model, the blood was collected from tail vein or by cardiac puncture to the left ventricle of the euthanized mice at day 26 post PolyI:C treatment. For the AAV-TBG-Cre; Mof f/f mouse model, the blood was collected from tail vein at day 26 post AAV injection. In all cases, serum was isolated by centrifugation. Analysis of the liver panel (AST, AKT, LDH, Cholesterol, Creatinine, Bilirubin, Triglycerides, ALK phosphatase) was performed at the University of Michigan Unit for Laboratory Animal Medicine (ULAM) Laboratory.

Preparation of the hepatocyte-conditioned medium
For hepatocyte-conditioned medium, primary hepatocytes isolated from the AAV-TBG-Cre; Mof f/f mice were plated at the density of 1.2 × 10 6 /ml in L15 media. The medium was collected after 24 h and stored at −80 C. For heat treatment, the conditioned medium was heated at 100 C for 5 min. For the L929conditioned medium, 2 × 10 5 L929 cells were seeded with 150 ml DMEM supplemented with 100 units/ml penicillin and streptomycin, 1% L-glutamine, and 10% FBS (Life Technologies, 10082). The medium was collected twice with 7-day interval and filtered (0.22 μM) before storage at −80 C.

Isolation and activation of bone marrow-derived macrophages (BMDMs)
Primary bone marrow cells were isolated from the femur and tibia of the Mof f/f ; ER-Cre mice and cultured in DMEM supplemented with 20% L929 conditioned medium, 10% heatinactivated FBS, 100 nM 4-OHT, 100 units/ml penicillin and streptomycin for 4 days to differentiate into bone marrowderived macrophages (BMDMs). BMDMs were activated by culturing with 50% hepatocyte-conditioned medium for 18 h. For coculture experiment, 1 × 10 5 BMDMs were seeded to an insert (Corning, CLS3428) with hepatocytes at bottom for 2 days before the experiment.

Mitochondria staining
Cells were grown on 12-mm coverslips and stained with 250 nM Mitotracker Red CMXRos (M7512, Molecular Probes) for 30 min at 37 C. The coverslips were washed twice with 1xPBS and three times with the growth medium. The cells were fixed with 4% formaldehyde for 15 min at room temperature (RT), permeabilized with 0.25% Triton X-100 for 30 min at 4 C, and stained with DAPI for 15 min at room temperature before microscopy.

RNA isolation, real-time quantitative PCR, and RNA-seq analysis
Total RNA was isolated from liver tissues, primary hepatocytes, or BMDMs using TRIzol (Invitrogen). A total 5 μg of RNA was reverse transcribed using SuperScript III Reverse Transcriptase (Invitrogen). Real-time quantitative PCR (RT-qPCR) was performed using Radiant Green 2x qPCR Mix Lo-ROX (Stellar Scientific) on a BioRad C1000 Touch ThermoCycler. Primers were listed in Table S4. All the statistical analysis for RT-qPCR was performed using GraphPad Prism 8 software. For RNA-seq analysis, triplicates of RNA were isolated, treated with DNase I. RNA integrity analysis was performed using an Agilent Bioanlyzer. Only RNA with RNA integrity numbers (RINs) of 8 or above was used to prepare libraries. The samples were sequenced on the Illumina HiSeq2000 platform. TopHat2 was used to map reads to mouse reference genome assembly mm9. Mapped reads were then analyzed by DESeq to identify differentially expressed genes. Gene expression was considered significantly different if 1) the adjusted p value was less than 0.05 and 2) log2 (fold change) was greater than 1 or less than −1. Volcano plots were generated using R software (http://www.rproject.org/). Clusters were identified using ClusterONE28 and analyzed for Gene Ontology (GO) terms using BinGO29. Unsupervised GO analysis was performed using all differentially expressed genes as input for DAVID (https://david.ncifcrf.gov) and visualized using GOPlot30 in R with false discovery rate (FDR) ≤ 0.05. KEGG pathway analysis was performed using upor downregulated gene sets in DAVID. Only pathways with adjusted p value ≤0.05 were considered significant.

Data availability
RNA-seq data for Mof f/f and Mof −/− liver tissues are deposited into NCBI's Gene Expression Omnibus (GEO) with accession number GSE106369. RNA-seq data for primary human NASH patient samples are downloaded from data set GSE134422 (49) and GSE126848 (50).